Quaternary Science Reviews 101 (2014) 28e35 Contents lists available at ScienceDirect Quaternary Science Reviews journal homepage: www.elsevier.com/locate/quascirev The demographic response to Holocene climate change in the Sahara * Katie Manning a, , Adrian Timpson a, b a Institute of Archaeology, University College London, 31-34 Gordon Square, London, WC1H 0PY, UK b Research Department of Genetics, Evolution and Environment, University College London, Darwin Building, Gower Street, London, WC1E 6BT, UK article info abstract Article history: The timing and development of Holocene human occupation in the now hyperarid Sahara has major Received 20 February 2014 implications for understanding links between climate change, demography and cultural adaptation. Here Received in revised form we use summed probability distributions from 3287 calibrated 14C dates from 1011 archaeological sites 24 June 2014 to demonstrate a major and rapid demographic shift between 10,500 and 5500 years BP. This event Accepted 1 July 2014 corresponds with the African Humid Period (AHP) and is sub-continental in scale, indicating climate as Available online the prime factor driving broad-scale population dynamics in northern Africa. Furthermore, by providing a high temporal resolution proxy for effective carrying capacity our population curve offers an inde- Keywords: African humid period pendent estimate of environmental change in northern Africa, indicating a temporal delay in the Sahara terrestrial response to atmospheric climate change. These results highlight the degree to which human Holocene demography is a function of environment at the appropriate scale of observation in both time and space Demography and sheds important new light on the social response to global environmental change. Human sustainability © 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/3.0/). 1. Introduction and Kropelin,€ 2006), meanwhile, have shown evidence for intensive occupation, initially by pottery-using hunter-gatherers, and then by The Sahara is the world's largest hot desert. With only 2.5 increasingly mobile pastoralists after the introduction of domestic million people for over 3.5 million square miles, it currently sup- livestock c. 7500 years BP (Marshall and Hildebrand, 2002; Dunne ports one of the lowest population densities on earth, yet this has et al., 2012). Taken together, these lines of evidence strongly sup- not always been the case. Over the course of the Holocene the port a Holocene occupation of the desert, indicating close links be- Sahara underwent major climatic changes, marking the beginning tween changing environments, human occupation and adaptation. of the African Humid Period (AHP) approximately 12,000 years ago However, although such regional perspectives provide vital infor- (Gasse et al., 1990; Street-Perrott and Perrott, 1990; deMenocal mation about the circumstantial influences of climate on Neolithic et al., 2000; Adkins et al., 2006; McGee et al., 2013; Tierney and populations, little is known about the broad scale population trends, deMenocal, 2013), and what has been referred to as the ‘Green or the degree of synchronicity between demographic change and Sahara’ (Sereno et al., 2008; Drake et al., 2011; Dunne et al., 2012). the onset and termination of the Holocene AHP. Whilst the Holocene AHP is perhaps one of the most thoroughly Here we present results on the demographic reconstruction of documented and well-dated climate change events, its implications the Saharan Holocene, revealing a temporal delay between the for ancient human settlement is less well understood. onset of humid conditions as represented in sedimentary dust flux Occupation is clearly testified in the frequent rock engravings records, and the human re-occupation of the desert. We estimate that are scattered throughout the upland regions of the desert, population growth and decline, and identify key stages in the illustrating a lush environment with Sahelian and riverine fauna and Sahara's recent demographic history; a massive increase in trans- scenes of large-game hunting, livestock herding and religious cer- Saharan population levels shortly after 11,000 years BP, a tempo- emony (Mori, 1965; di Lernia and Gallinaro, 2010). Settlement rary decline in population levels between 7600 and 6700 years BP, studies from the Fezzan (Cremaschi and Zerboni, 2009; Cancellieri and a major population collapse at the end of the AHP, between and di Lernia, 2013), Tenere (Sereno et al., 2008; Garcea, 2013) and 6300 and 5200 years BP. Eastern Sahara (Wendorf et al., 2001; Hoelzmann et al., 2002; Kuper 1.1. Timing and abruptness of the Holocene African Humid Period * Corresponding author. Tel.: þ44 207 679 1530. Debate is on going over the timing and abruptness of the AHP. E-mail address: [email protected] (K. Manning). Both sediment flux records from deep sea cores off the coast of http://dx.doi.org/10.1016/j.quascirev.2014.07.003 0277-3791/© 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/3.0/). K. Manning, A. Timpson / Quaternary Science Reviews 101 (2014) 28e35 29 Africa. Full details on the method can be found in Shennan et al. (2013) and Timpson et al. (in press), and all dates are calibrated using the IntCal09 curve (Reimer et al., 2009). Our initial analysis uses 3287 published radiocarbon dates from 1011 Neolithic archaeological sites (Fig. 1). Further details about the data and the full radiocarbon date list can be found in Appendix A. 2.1. Bootstrap analysis We have also used a bootstrap analysis to provide an additional broad indication of the extent to which the Summed Probability Distribution (SPD) shape can be attributed to a genuine underlying pattern in the data, rather than random sampling fluctuations from Fig. 1. Map of the study area showing all 1011 sites included in the analysis. The four a small sample. We generated 1000 bootstrapped SPDs by resam- sub-regions are distinguished by colour. Red represents the Eastern Sahara, blue the pling (with replacement) the observed dates and calculated local Central Sahara, orange the Atlas and purple the Western littoral. The grey dots 95% confidence intervals to indicate regions with greater represent sites that were not included in any of the four sub-regional analyses, but confidence. were included in the overall analysis. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) 2.2. Calculating the effects of climate on human demography Initially we attempted a regression analysis on our demographic north-west Africa (deMenocal et al., 2000; Adkins et al., 2006; curve and several high-resolution palaeoclimate proxies derived McGee et al., 2013) and recent correlation analysis of dDwax iso- from sedimentary dust flux records (Adkins et al., 2006). However, topic values from east and northeast Africa (Tierney and as shown in Section 3.1 and 3.2, and discussed below, formal cor- deMenocal, 2013) point to a rapid hydroclimatic shift between relations were problematic. In order to test the hypothesis that 12,500e11,500 and 5500 years BP. Pollen and sedimentological climate change forced major population changes in the Saharan records from Lake Yoa, in northern Chad however, indicate a more Holocene, we therefore divided our global dataset into four broad gradual decline in precipitation and subsequent progressive drying regions (Fig. 1), the Atlas, Eastern Sahara, Central Sahara and the of the regional ecosystem (Kropelin€ et al., 2008; Francus et al., Western littoral and tested for correlation between the SPDs for 2013). This discrepancy is as yet unresolved, and is most likely a these regions. The rationale behind this approach is that if pop- result of the differential sensitivity of the various proxies ulations respond concurrently across a diverse and large enough (Holelzmann et al., 2004; Francus et al., 2013; Tierney and area, we can infer an exogenous cause such as climate to be the deMenocal, 2013), although changes in regional hydroclimates driving force. These sub-regions are purposefully very broad in may also have been amplified by vegetation feedback (Renssen order to maintain statistically viable sample sizes. Nonetheless, the et al., 2006), and local groundwater conditions (Holelzmann boundaries between sub-regions are established in light of general et al., 2004; Lezine et al., 2011; Hely et al., 2014). Patches of more geographic features, such as the distribution of palaeohydrological arid terrain, for example, are hypothesized for parts of the Central catchment areas, and to a lesser extent, cultural features such as the Sahara based on pollen core records from the North Atlantic distinct pottery styles of the Western Littoral. Correlation co- (Hooghiemstra et al., 1992) and Holocene lakes in the Central efficients between each of the four sub-regions are calculated using Sahara (Hely et al., 2014) as well as the fact that some sub-Saharan Pearson's R. However, a certain amount of similarity can be ex- aquatic species do not exhibit trans-Saharan spatial distributions pected due to short-term fluctuations inherited from the calibra- (Drake et al., 2011). A key issue therefore, is to assess the chro- tion process, since all data is calibrated using the same calibration nology of demographic change, and the spatial range over which curve. Therefore p-values are estimated by comparing the corre- those adaptations occurred. lation between the observed SPDs with the correlation between simulated SPDs. To ensure p-values are even more conservative, 2. Material and methods simulations are generated using a uniform null model, which fits the observed trends closer than exponential. Five hundred simu- In this paper we use a Monte-Carlo Summed Probability Dis- lated SPDs are generated for each sub-region (as detailed in tribution (MCSPD) method, which assesses the combined proba- Shennan et al., 2013) and for each pair of sub-regions the correla- bility distribution of a large number of radiocarbon dates.